Embed Size (px)
Citation preview
General rights Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.
Users may download and print one copy of any publication from the public portal for the purpose of private study or research.
You may not further distribute the material or use it for any profit-making activity or commercial gain
You may freely distribute the URL identifying the publication in the public portal If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.
Downloaded from orbit.dtu.dk on: Oct 05, 2021
Evidence of spatiotemporal variations in contaminants discharging to a periurbanstream
Lemaire, Grégory Guillaume; McKnight, Ursula S.; Schulz, Hanna; Roost, Sandra; Bjerg, Poul Løgstrup
Published in:Ground Water Monitoring & Remediation
Link to article, DOI:10.1111/gwmr.12371
Publication date:2020
Document VersionPeer reviewed version
Link back to DTU Orbit
Citation (APA):Lemaire, G. G., McKnight, U. S., Schulz, H., Roost, S., & Bjerg, P. L. (2020). Evidence of spatiotemporalvariations in contaminants discharging to a periurban stream. Ground Water Monitoring & Remediation, 40(2),40-51. https://doi.org/10.1111/gwmr.12371
1
Evidence of spatio-temporal variations in contaminants discharging to a peri-urban 1
stream 2
Gregory G. Lemaire1*, Ursula S. McKnight1, Hanna Schulz1,2, Sandra Roost3, Poul L. Bjerg1 3
1 Department of Environmental Engineering, Technical University of Denmark, Bygningstorvet 115, 2800 4
Kgs. Lyngby, Denmark 5
2 Leibniz-Institute of Freshwater Ecology and Inland Fisheries, Müggelseedamm 310, 12587 Berlin, Germany 6
(present address) 7
3 Orbicon, Linnés Allé 2, 2630 Taastrup, Denmark 8
*Corresponding author: [email protected] 9
10
2
ABSTRACT 11
Xenobiotic organic compounds can be discharged from contaminated groundwater inflow and may seep 12
into streams from multiple pathways with very different dynamics, some not fully understood. In this study, 13
we investigated the spatio-temporal variation of chlorinated ethenes discharging from a former industrial 14
site (with two main contaminant sources, A and B) into a stream system in a heterogeneous clay till setting 15
in eastern Denmark. The investigated reach and near-stream surroundings are representative of peri-urban 16
settings, with a mix of high channel alteration and more natural stream environment. We therefore 17
propose an approach for risk assessing impacts arising from such complex contamination patterns, 18
accounting for potential spatio-temporal fluctuations and presence of multiple pathways. Our study 19
revealed substantial variations in pathway contributions and overall contaminant mass discharge to the 20
stream. Variable contaminant contributions arising from both groundwater seepage and urban drains were 21
identified in the channelized part of the north stream, primarily from source A. Furthermore, variations in 22
the hyporheic and shallow groundwater flows were found to enhance contaminant transport from source 23
B. These processes result in an increase of the overall mass of contaminant discharged, correlating with the 24
channels’ flow. Thus, an in-stream control plane approach was found to be an effective method for 25
integrating multiple and variable discharge contributions quantitatively, although information on specific 26
contaminant sources is lost. This study highlights the complexity and variability of contaminant fluxes 27
occurring at the interface between groundwater and peri-urban streams, and calls for the consideration of 28
these variations when designing monitoring programmes and remedial actions for contaminated sites with 29
the potential to impact streams. 30
Keywords: Risk assessment, contaminant mass discharge, chlorinated ethenes, groundwater-surface water 31
interactions, spatio-temporal variation, peri-urban streams 32
33
34
3
1 Introduction 35
The implementation of legislations such as the European Water Framework Directive, which aim to protect 36
water resources and associated ecosystems, are challenged by extensive anthropogenic activities 37
worldwide (McKnight et al. 2012; Vörösmarty et al. 2010). Freshwater resources are particularly sensitive 38
and the related ecosystems exhibit some of the highest decline rates in terms of biodiversity (Strayer and 39
Dudgeon 2010; Sonne et al. 2018). The reasons for such a decline are many and possibly cumulative, 40
including stream flow alteration, habitat degradation, pollution, invasive species and more recently 41
emerging contaminants and climate change (Reid et al. 2018; Sonne et al. 2017). 42
The rapid urbanization observed worldwide has led to a fast and sometimes chaotic development of areas 43
located at the fringe of existing cities or rural and secondary towns. This evolution has contributed to the 44
development of a heterogeneous patchwork of urban, rural and natural areas now defined as peri-urban 45
‘scapes’ (Eakin, Lerner, and Murtinho 2010; Simon 2008; Allen 2003). These new “landscapes” and resulting 46
mixtures of activities are having profound impacts on the surrounding freshwater, especially for streams 47
and rivers draining these types of catchments. Importantly, stream flow regimes may be modified by the 48
coexistence of mixed catchment response times stemming from the more impervious urban-drained areas 49
and the more rural/natural-drained areas, with possible reductions in evapotranspiration and base flow 50
(Simon 2008; Braud, Fletcher, and Andrieu 2013; Walsh et al. 2016). Furthermore, water quality is often 51
directly affected by the rise of urbanization, with possible increases in water temperature, suspended 52
solids, and enhanced loads of chemical pollutants that may discharge via numerous pathways leading to 53
counterintuitive patterns in contaminant distributions (Walsh et al. 2016; Simon 2008; Paul and Meyer 54
2001; Zoboli et al. 2019). 55
Due to their persistency, ubiquity and low degradation potential, chlorinated aliphatic hydrocarbons (CAH) 56
such as trichloroethene (TCE) remain a major concern regarding legacy contaminants in groundwater for 57
many industrialized countries (Rivett et al. 2012; Freitas et al. 2015; Ellis and Rivett 2007; Conant et al. 58
4
2004). Furthermore, the mobility of these compounds and their associated metabolites, such as cis-59
dichloroethene (cDCE) and vinyl chloride (VC), make them a serious threat to receiving stream waters 60
(McKnight et al. 2010; Freitas et al. 2015; Weatherill et al. 2018; Rønde et al. 2017). Quantification of CAH 61
fluxes in groundwater from former industrialized and contaminated sites can be carried out using the 62
contaminant mass discharge (CMD) approach, where the mass per unit time flowing through a virtual 63
control plane is quantified (Basu et al. 2006; Weatherill et al. 2014). This approach has been successfully 64
applied to the characterization of contaminant plumes in aquifers (Weatherill et al. 2014), although some 65
inherent uncertainty exists (Rønde et al. 2017; Troldborg et al. 2010). This methodology has also been 66
employed to track the fate of contaminants in streams (in-stream CMD), with a simplified approach 67
assuming a full mixing of the contaminants (Rønde et al. 2017; Milosevic et al. 2012; LaSage et al. 2008). 68
Numerous studies have shown the importance of the groundwater pathway for the discharge of CAH to 69
streams (Chapman et al. 2007; Conant et al., 2004; Ellis & Rivett, 2007; Paul & Meyer, 2001; Rasmussen et 70
al., 2016; Roy et al., 2016; Roy and Bickerton, 2012; Shepherd et al., 2006; Sonne et al., 2017). However, 71
the transport pathways for contaminants arising from urban and peri-urban areas may be even more 72
complex and result in variable release dynamics for which little is known (Ellis and Rivett 2007; Braud et al 73
2013). So although contaminants may reach the stream via groundwater-surface water interactions, there 74
may be more rapid pathways triggered simultaneously from dense underground infrastructures (e.g. pipes 75
or trenches) in more urban-like settings (Kaushal and Belt 2012; Hatt et al. 2004). Notably, a few studies 76
apply the in-stream CMD methodology at several points in time to investigate the groundwater 77
contribution and fate of CAHs (and other dissolved contaminants) in streams (Heejung and Hemond, 1998; 78
LaSage et al., 2008a; Milosevic et al., 2012; Rønde et al., 2017). Quite often, the contaminant mass balance 79
could not be completely closed, indicating the existence of additional pathways for which the dynamics 80
were not quantified at the time of the study. 81
This lack of knowledge concerning the contaminant dynamics is a shortcoming in the assessment of the 82
chemical status of streams. Monitoring programs traditionally focus on the measurement of stream 83
5
concentrations for a direct comparison with the relevant Environmental Quality Standards (EQS) at regular 84
time intervals (see, for example, European Commission (2009)). High frequency sampling would be ideal 85
(Brack et al. 2017; Skeffington et al. 2015), but is rarely feasible predominantly due to economic 86
constraints. This focus on specific measurement points in the stream sampled on a regular basis may be 87
misleading and the risk associated to a given pollutant not properly assessed. In a recent study, Zoboli et al. 88
(2019) captured increasing micropollutant concentrations in different Austrian river catchments despite 89
higher stream flows and resulting dilutions. This “counterintuitive” pattern was due to the increased 90
contribution of multiple sources during higher flow regimes, counterbalancing the positive effect of the 91
enhanced dilution. In other words, the dynamic of the sources itself was driving the general trend of the 92
measured concentrations. Thus, improving our understanding of the dynamics of contaminant source 93
releases, associated discharges and dominant pathways would benefit both the design and implementation 94
of monitoring programs, as well as the assessment of chemical status (Sonne et al. 2017). 95
The objectives of this paper are therefore to investigate the spatial variation and discharge dynamics 96
originating from a contaminated site impacting a peri-urban stream in order to (1) quantify the potential 97
temporal variations on contaminant mass discharge to the stream, and resulting stream concentrations; (2) 98
address or identify the key processes and pathways responsible for such variations; and (3) discuss the 99
findings at the site in relation to the development of key legislations, and monitoring strategies. To do so, 100
we applied a discrete in-stream contaminant mass discharge approach at relevant transects along the 101
investigated stretch. We argue that this method allows accounting for the non-fully mixed conditions 102
occurring when multiple contaminant pathways are active, and is crucial when designing suitable 103
monitoring programs capable of capturing the impacts of contaminated sites and resulting chemical status 104
in streams. 105
106
2 Study site 107
6
2.1 Site description 108
The study site is a stream reach ca. 500 m long called the Mølleåen, located 15 km north of Copenhagen in 109
the town of Raadvad (Fig. 1). Raadvad is the site of a former industrial area, primarily used over the past 110
two centuries for metal manufacturing activities, and is the location for a well-known knife factory that 111
produced domestic metal wares until 1972. In addition to metal works, the site has hosted a number of 112
other factories including a bronze foundry and metal workshop. The site today houses various professions 113
including offices, residential houses and a Nature school offering a variety of activities for children.114
. 115
Figure 1. Overview of the Raadvad site showing land use, stream flow direction and location of the measurement transects. The 116
dashed red lines correspond to transect locations where detailed stream water sampling was carried out (5 points per transect); 117
solid red lines show the transects used for the in-stream contaminant mass discharge estimation. The dashed black line box 118
indicates the location of the north channel presented in more detail in Figure 2. 119
7
The annual precipitation in the area is ca. 860 mm (St. 5628; Danish Meteorological Institute, 2017). The 120
stream flows eastwards with a depth ranging from 0.3 to 1.5 m and width from 6 to 12 m along the 121
investigated stretch. The median stream flow is ca. 437 l/s measured at a monitoring station 500 m 122
upstream of the investigated site for the period 2007-2018 (St. 5010; Danmarks Miljøportal, 2018). The 123
urban part of the stream reach consists predominantly of two engineered channels running past the 124
industrial site. The channels flow from an impounded lake through two low-head dams that are rarely 125
operated (i.e. remain in a fixed position), located just west (upstream) of the site. The north channel (the 126
original location of the stream prior to industrialization) is straight and was initially channelized for a water 127
mill used by the industry in the 1900s. It is bounded by a concrete retaining wall on the northern bank in its 128
most upstream section, with a streambed consisting of organic sediments cluttered with rocks. The 129
southern channel flows through a more natural setting and the streambed consists of pebble and stone 130
with patches of organic sediments. These channels then merge to reform the Mølleåen downstream of the 131
industrial area and flow through a protected nature area into the Baltic Sea. 132
2.2 Geology and hydrogeology 133
The geology in the area of interest is typical for Eastern Denmark. The setting is complex, where gyttja 134
dominates in the vicinity of the streambed, and below this a mixture of layers of sand, gravel and clay till 135
forms a shallow heterogeneous aquifer. Below this shallow aquifer, and separating this layer from a deep 136
Danien limestone aquifer, is another more continuous clay till layer. The terrain is relatively flat at an 137
average elevation of 5 m.a.s.l. (DVR90) with a slight slope from the north side, leading to a shallow 138
groundwater flow towards the stream. It is not known to what extent the shallow aquifer and surface 139
water are hydrologically connected. Previous investigations reported no indication of separate phase 140
mobile CAHs, upward hydraulic gradients from the deep to shallow aquifers, as well as from the shallow 141
aquifer to the stream (Niras 2012). This ruled out contamination (see next section) of the deep aquifer, 142
which is not discussed further in this paper. 143
8
2.3 Contamination overview 144
Previous investigations of the soil (pore air) and groundwater revealed a contamination of the shallow 145
aquifer by CAH (specifically, TCE, cDCE, VC), as well as petroleum hydrocarbons from the leakage of 146
underground and surface storage tanks, urban drains and likely waste deposits placed randomly 147
throughout the site (Niras 2012, Orbicon 2016). Contamination by heavy metals and oil in the topsoil and 148
stream sediments were also detected over most of the industrial site. Figure 2 depicts the locations of the 149
main contaminant source zones for CAH (termed source zone A and B throughout the paper) impacting the 150
shallow groundwater: source zone A is located below the main building on the northern bank of the 151
northern channel, and source zone B is located at the junction of the two channels on the southern bank of 152
the north channel. Finally, a third and comparatively minor source area was also detected in the most 153
upstream part of the south channel, releasing mostly TCE to the stream. However, the concentration levels 154
were not substantial compared to the contribution of the two other sources (Niras 2012; Orbicon 2016; 155
Lemaire 2016), and was therefore not investigated further in this study. 156
Previous investigations by consulting companies revealed CAH concentrations of up to ca. 2500 µg/l in the 157
shallow groundwater, at a monitoring well 1 m from the north bank (see location in Fig. 2). Some of the 158
contamination present in source zone A was further found to be entering the stream, as confirmed by grab 159
samples in the vicinity of the northern bank, and upstream and downstream of the retainment wall (Niras 160
2012, Orbicon 2016). The exact seepage location and contaminant pathways were, however, unknown. 161
Preferential pathways were suspected from former urban drains whose outlets are visible in the retaining 162
wall, as well as through some cracks and fissures made visible by the presence of precipitated iron oxides 163
(see pictures of these features in Fig. 2). Total CAHs of up to 2600 µg/l were measured in the shallow 164
groundwater in source zone B (Niras 2012). However, its potential discharge to the stream had not been 165
estimated prior to this study. 166
9
167
Figure 2: (a) Locations for stream water (blue circles) and streambed (piezometers; open squares) samples used in the estimation 168
of contaminant mass discharge (CMD). Approximate delineation (sum of CAHs concentration > 10 µg/L in shallow groundwater) 169
for the main contaminant source zones A and B are marked with striped red polygons based on previous investigations by a 170
consulting company (Niras 2012). Note that the other transects used for CMD estimation are shown in Fig. 1. (b) Pictures from 171
the site indicate the location of the crack and urban drain in the retainment wall. 172
173
174
10
175
3 Materials and methods 176
3.1 Overview and approach to investigations 177
A one-year measurement campaign was carried out at the investigated site from June 2016 to May 2017, 178
focusing on the variations of contaminant mass discharge to the stream from the two contaminant sources. 179
The measurements encompassed groundwater, streambed and surface water sampling, flow discharge 180
measurements, as well as groundwater table and stream water level monitoring (10 points) repeated on a 181
monthly basis. A preliminary conceptual model was developed using extra stream water samples (in 182
addition to the first set of measurements collected), which consisted of 5 points along each of the first 5 183
transects (T1 to T5, Fig. 1) in order to document the mixing pattern of the CAHs in the stream water. This 184
was supported by application of a 2D advection-dispersion model of pollutant mixing in streams (Aisopou 185
et al., 2015; Lemaire, 2016) to evaluate the potential location of the discharge to the stream from source 186
zone A (i.e. with respect to chainage, streambed or streambank seepage), and to estimate the CMD and 187
mixing conditions. These results were then used to design all other monitoring campaigns: these consisted 188
of 3 points per transect for transects T1 and T4, and 1 point in transect T6 and T7 where fully mixed 189
conditions (see section 3.5) were achieved (see Fig. 1 for all transect locations). Finally, to better 190
understand some of the variations observed in our dataset, new streambed and stream water samples 191
were collected in June and November 2017 in the north channel, and additional monitoring of the stream 192
water levels within this area including the south channel was performed from October to December 2018. 193
3.2 Water level monitoring and stream discharge 194
The stream discharges in the north and south channel were obtained by monthly measurements using an 195
OTT MF pro flow meter and assessed using the mid-section method following the ISO 748 standard. The 196
stream flow measurements were taken at transects T1 and T7 for the north and south channel respectively 197
(Fig. 1 and 2). The flow rate in Mølleåen could not always be measured due to the water depth in periods of 198
11
high flow. Consequently, the flow rate was assumed equal to the summation of the north and south 199
channel flows, i.e. the possible additional groundwater influx was assumed minor over the limited length of 200
the investigated reach. 201
Hydraulic heads in the shallow aquifer and the stream water level were monitored hourly during the entire 202
campaign using mini divers (DI501) from Van Essen. The shallow groundwater was monitored in the 203
existing boreholes in source zone A (screened from 0.5 to 2.5 m.b.s) and B (screened from 1 to 4 m.b.s). 204
The stream water level was monitored approximately 10 m downstream of transect T1. 205
206
12
207
3.3 Hyporheic zone (streambed) sampling 208
Streambed water samples were collected using drive-point piezometers, placed within the hyporheic zone 209
to a depth of 40 cm in the most upstream part of the northern channel. The samples were extracted by use 210
of a peristaltic pump and rigid polyamide tubes, stored in 25 ml glass vials, sealed without air bubbles. The 211
number of drive-points varied across the 12 campaigns in order to enable the best delineation of the 212
contamination entering the stream via the streambed. The measurement locations were chosen based on 213
our conceptual model of the site, inspection of the streambed and ensuing practical constraints (stones, 214
rocks and obstacles in the streambed itself). The piezometers were purged 3 times prior to sampling to 215
ensure a freshwater input without the influence of stagnation. The samples were kept in a cooler on ice 216
during the sampling, and thereafter kept at 4⁰C prior to chemical analysis. All chemical analyses for CAH 217
were carried out by an independent certified laboratory using P&T/GC-MS method according to ISO 15680. 218
Detection limits for the analyses are 0.02 μg/L. 219
3.4 Stream water sampling 220
Stream water grab samples were collected in 25 ml glass vials, sealed without air bubbles. The locations 221
and numbers of samples for each of the measurement transects were previously described in section 3.1). 222
Stream water was sampled from the most downstream point to the most upstream to avoid any sediment 223
disturbance or cross contamination while measuring. They were kept on ice in a cooler on site before being 224
stored at 4⁰C until sent for analysis. An additional transect, T0, was placed up-gradient of transect T1 in 225
April, June and November 2017 to enable delineation of contamination originating from the urban drainage 226
system (see Fig. 4). 227
3.5 In-stream contaminant mass discharge 228
The CAH concentrations measured in the stream are dependent on the stream flow, diluting any 229
contaminants entering the stream. The temporal or spatial variation and fate of the discharged 230
13
contaminant mass can therefore be evaluated by assessing the mass flowing through different control 231
transects along the stream. Several formulations and simplifications can be used depending on the location 232
of this control transect with respect to the different points of discharge. 233
The in-stream concentration is generally dependent on both the transverse and depth location of the 234
sample within a control transect. For most streams, however, the depth is usually small compared to the 235
width and the vertical mixing occurs over a relatively short distance (Fisher 1979; Aisopou et al. 2015). At a 236
certain distance downstream of the discharge location, the contaminant will therefore be completely mixed 237
in the vertical direction and the contaminant concentration is then only dependent on the transverse 238
location of the sample. Thus the in-stream CMD, 𝐽𝐽 can be evaluated using a discretization of the 239
measurements taken along the in-stream control plane (ITRC 2010): 240
241
𝐽𝐽 = �𝑐𝑐𝑖𝑖𝐴𝐴𝑖𝑖𝑞𝑞𝑖𝑖
𝑛𝑛
𝑖𝑖=1
(1)
where 𝑐𝑐𝑖𝑖 is the pollutant concentration at mid water column in the sub-area 𝑖𝑖, 𝑞𝑞𝑖𝑖 is the flow velocity in the 242
sub-area of the in-stream control plane of area 𝐴𝐴𝑖𝑖 and 𝑛𝑛 is the number of sub-areas. In our study, n=3 243
corresponding to the 3 water samples taken at regularly spaced locations along each of the transects T𝑖𝑖, 𝑞𝑞𝑖𝑖 244
estimated as the average velocity in this sub-area, and 𝐴𝐴𝑖𝑖 computed using the average depth of the sub-245
area 𝑖𝑖 and width. 246
A contaminant is considered to be fully mixed when its concentration along a transverse section of the 247
stream remains within 5% of the average concentration (Fisher, 1979). In that case, the concentration is 248
almost constant in the control transects, both in depth and transverse direction. Hence the formula (1) 249
simplifies and can be rewritten as (Rønde et al. 2017): 250
𝐽𝐽 = 𝑐𝑐𝑚𝑚𝑖𝑖𝑚𝑚 ∗ 𝑄𝑄𝑚𝑚𝑖𝑖𝑚𝑚 (2)
14
where 𝑐𝑐𝑚𝑚𝑖𝑖𝑚𝑚 is the fully mixed concentration and 𝑄𝑄𝑚𝑚𝑖𝑖𝑚𝑚 the stream flow rate at the point of fully mixed 251
conditions downstream. Equation (2) is hereby applied to estimate the contaminant mass discharge at 252
transect T6 and T7. 253
Finally, as parent and degradation products have been measured (i.e. TCE, cDCE, VC), the CMD results are 254
reported using the PCE equivalent unit, defined as: 255
𝑚𝑚𝑃𝑃𝑃𝑃𝑃𝑃,𝑒𝑒𝑒𝑒 = 𝑀𝑀𝑃𝑃𝑃𝑃𝑃𝑃,𝑒𝑒𝑒𝑒�𝑚𝑚𝑐𝑐ℎ𝑙𝑙. 𝑐𝑐𝑐𝑐𝑚𝑚𝑐𝑐𝑐𝑐𝑐𝑐𝑛𝑛𝑐𝑐
𝑀𝑀𝑐𝑐ℎ𝑙𝑙. 𝑐𝑐𝑐𝑐𝑚𝑚𝑐𝑐𝑐𝑐𝑐𝑐𝑛𝑛𝑐𝑐
(3)
where 𝑚𝑚𝑖𝑖 is the mass of a given chlorinated compound or metabolite 𝑖𝑖, and 𝑀𝑀𝑖𝑖 is its corresponding molar 256
mass. 257
3.6 Time series and statistical analysis 258
The investigation of time variations and interactions between the different parameters of interest 259
(precipitation, hydraulic head, contaminant mass discharge, etc.) was carried out using Spearman rank and 260
cross-correlation analyses. The cross-correlation analysis was employed to estimate an average lag 261
between the input driving parameter (i.e. precipitation) and output hydrological responses defined by the 262
different hydraulic head measurements in the stream and shallow groundwater, following the techniques 263
described in (Larocque et al.,1998). The discrete cross-correlation function is defined as: 264
𝜌𝜌𝑋𝑋𝑋𝑋(𝑘𝑘) =𝐶𝐶𝑋𝑋𝑋𝑋(𝑘𝑘)𝜎𝜎𝑋𝑋𝜎𝜎𝑦𝑦
(4)
265
with: 266
𝐶𝐶𝑋𝑋𝑋𝑋(𝑘𝑘) = 1𝑛𝑛�(𝑥𝑥𝑡𝑡 − �̅�𝑥)(𝑦𝑦𝑡𝑡 − 𝑦𝑦�)𝑛𝑛−𝑘𝑘
𝑡𝑡=1
(5)
15
267
where �̅�𝑥 and 𝑦𝑦� are the means of time series 𝑥𝑥 and 𝑦𝑦, 𝜎𝜎𝑋𝑋,𝜎𝜎𝑦𝑦 are the respective standard deviations and 𝑛𝑛 is 268
the length of the time series. 269
A Spearman rank correlation analysis was then carried out to highlight potential trends between the 270
different time series, especially as these trends are likely to be non-linear and the data not normally 271
distributed. In order to do so, the hourly hydraulic head time series datasets were down-sampled to the 272
measurement frequency (once a month) by averaging over the 4-hr intervals during which the stream 273
water samples were collected. The hourly time series for the precipitation data was down-sampled to the 274
measurement frequency by a cumulative estimate of precipitation between the previously estimated lag 275
(determined by the cross-correlation) and the end of the measurement intervals to account for the time 276
response of the hydrological system. 277
278
16
279
4 Results and Discussion 280
4.1 Spatial distribution of the contaminants in the stream 281
The first sampling campaign, carried out in June 2016, revealed a contamination of the stream water by TCE 282
and its degradation products cDCE and VC (Fig. 3). cDCE was the dominant CAH compound in terms of 283
concentration levels along the investigated stream stretch. Specifically, cDCE concentrations of up to ca. 2 284
µg/L were measured in the north channel close to the northern bank in the most upstream transect, T1, 285
where source zone A is discharging. After the junction of the two channels, the concentrations decreased 286
significantly due to dilution resulting from the flow coming from the south channel. Figure 3 furthermore 287
indicates the pattern of mixing across the different transects, where concentration values “flatten out” with 288
distance due to the dispersion and spreading of the contaminants in the transverse direction. However, 289
fully mixed conditions have not been attained even at the most downstream transect location, T5, for this 290
specific campaign based on the criteria defined in section 3.5. 291
Taken altogether, this indicated that the bulk of the contamination originates from the northern channel, is 292
heavily diluted by water inflowing from the south channel, and that the location of fully mixed conditions 293
requires additional measurements further downstream of transect T5. 294
17
295
Figure 3. Effects of mixing and dilution of the in-stream contaminant concentrations measured at 5 detailed transects for TCE 296
and detectable degradation products (cDCE and VC) in June 2016. Note that the x-axis corresponds to a normalized stream 297
width, i.e. 0 for the north bank, and 1 for the south bank. 298
4.2 Investigating contaminant pathways from source zone A into the stream 299
Differences in hydraulic head between the streambed piezometers installed in the north channel and the 300
stream water level were found to vary in time, but always upwards flow (i.e. flow direction from the 301
streambed to stream) or null at the time of measurements, and on the order of a few mm to a maximum of 302
4 cm. The corresponding water samples were also variable, revealing a highly heterogeneous overall 303
contamination pattern, with concentrations of CAHs varying, for example in April 2017, from a few µg/L up 304
to ca. 2,600 µg/L (PCE eq) over a small confined area close to the retaining wall (Fig. 4a). The 305
18
measurements taken from the monitoring well in the shallow aquifer revealed concentrations ranging from 306
ca. 4300 to 7500 µg/L (PCE eq) for all sampling campaigns. 307
308
Figure 4. Variations of chlorinated ethene concentrations in the groundwater monitoring well, streambed, and stream in the 309
north channel. a) Spatial variations for a selected period (April 2017) expressed in PCE eq. ; b) Temporal variations of molar 310
ratios for the stream water and streambed water samples at different locations for three selected periods (April, June and 311
November 2017). 312
19
Notably, the molar ratios between the different chlorinated compounds were stable in time for both the 313
shallow aquifer and hyporheic zone samples: VC was almost exclusively found under the streambed and 314
mostly cDCE in the shallow aquifer (Fig. 4b). However, significant temporal variations in the molar ratios for 315
cDCE and VC were observed in the stream water samples compared to the relatively stable ratios observed 316
in the shallow aquifer. These temporal variations were especially noticeable directly up- and downstream 317
of the crack in the retaining wall, as well as in the ‘temporary’ transect (T0), and in the recess and drain (Fig. 318
4b). 319
These observations suggest that the contaminants discharging into the north channel from source zone A 320
occurred via different pathways with variable contributions. VC mostly discharged via the streambed, 321
indicative for a groundwater flow pathway, while the cDCE component entered via different hydrological 322
preferential flow paths from the northern bank (cracks and urban drains). These discharges were extremely 323
dynamic and their respective contribution dependent on the near-surface hydrological flow conditions and 324
hydrogeological properties of the stream bottom. These results are thus reflective of a highly complex 325
system, comprised of natural discharge processes combined with urban flow paths, leading to a highly 326
complicated and heterogeneous contaminant discharge pattern that varies in space and time over a 327
relatively limited area. Nevertheless, all contaminant pathways ultimately converge downstream of this 328
complex discharge zone and add up with the additional CMD from source zone B. Currently the Danish 329
authorities requires an understanding of the total impact from each contaminated site impacting surface 330
waters. From there, the focus of our study shifted from this single source area to determine the overall 331
CMD variation originating from the site. 332
4.3 Temporal variation of the in-stream contaminant discharge 333
The temporal variations of the in-stream mass discharge for CAH over 12 months were evaluated at 334
transects T1, T4 and T6 (Fig. 5). When considered individually, significant temporal variations occurred and 335
increased from T1 to T6 with an average yearly discharge of 1.7, 4.7, and 3.7 kg/yr PCE eq (coefficient of 336
20
variation, CV = 67%, 75% and 96%, respectively). In transect T1, downstream source zone A, the highest 337
estimate for in-stream contaminant discharge was observed in January and February 2017, while the most 338
important discharges after the channel junction at transect T4 were found to not coincide with the ones in 339
T1. 340
341
342
Figure 5: Temporal variations of in-stream contaminant mass discharge for chlorinated compounds (TCE, cDCE and VC) along the 343
north channel (T1) and downstream transects (T4, T6), and in the south channel (T7), expressed as PCE eq. See section 3.5 for 344
calculations. 345
21
At this particular transect, T4, a significant increase in the in-stream CMD was observed for three of the 346
measurement rounds including November 2016 (ranging e.g. from 1.9 in T1 to 13.4 kg/y PCE eq in T4), as 347
well as February and March 2017. These elevated levels in CMD held in “intensity” further downstream to 348
T6 for both November 2016 and March 2017; while continuously increasing in April 2017. This 349
intensification of contaminant mass discharge to the stream was not caused by a sudden discharge from 350
the south channel, considering the relatively low contaminant mass that has been estimated for that point 351
in the same periods (see T7, Fig. 5). Instead, a significant CAH discharge must have been active during these 352
periods and under specific hydrological conditions occurring between transects T1 and T4. The attention 353
was then shifted to assessing the role of source zone B (Fig. 2) and the potential processes governing the 354
increase in CMD. 355
4.4 Evaluation of governing environmental processes 356
Spearman rank correlation was used to evaluate the interrelationships between precipitation, monitored 357
hydrological parameters and estimated in-stream CMD in order to investigate the potential drivers for the 358
contaminant dynamics found downstream of the junction at transect T4. Correlations were in fact found to 359
be significant for describing the variation of in-stream CMD with some of the parameters collected on site 360
(see Fig. 6). Specifically, the in-stream CMD, 𝐽𝐽, for both cDCE and VC exhibited high positive correlation 361
values with the water flow rate, Q, for both the north and south channels (𝜌𝜌𝑠𝑠> 0.6 with p< 0.05, Figure 6a). 362
The in-stream CMD at transect T4 varied almost linearly with the measured flow rate in the north channel 363
(R2 > 0.8 for both cDCE and VC, Fig. 6b), and to a certain extent with the measured flow rate in the south 364
channel (R2 > 0.3, with one outlier removed corresponding to an unusually high stream flow in the south 365
channel only). 366
No significant Spearman rank correlations (Fig. 6a, 𝜌𝜌𝑠𝑠< 0.6 and corresponding p> 0.05) were observed 367
between the in-stream CMD estimated at transect T4, and drivers/proxies of shallow groundwater flow 368
such as precipitation, N, or the relative water level gradient, Δh, measured between the north channel and 369
22
the shallow groundwater monitoring well in source zone A. We therefore speculate that the increase in 370
CAH mass at transect T4 is not caused by a shallow groundwater pathway not accounted for stemming 371
from source zone A. Instead, the interaction between the two channels and source zone B was suspected as 372
the main driver responsible for this mass increase. We also ruled out the possibility of a resuspension of 373
sorbed contaminants, considering the significant increase of CAH between T1 and T4 (ca. 7 times more in 374
November 2016). 375
376
Figure 6. (a) Spearman rank correlation, 𝝆𝝆𝒔𝒔, diagonal matrix between in-stream contaminant mass discharge in transect T4 and 377
selected hydrological parameters: precipitation (N), stream flow rates (Q), water table in shallow aquifer (hGW) at source zone A 378
and B, as well as stream water table (hSW) and the hydraulic gradient (Δh). Subscripts A and B refer to source zones A and B, 379
respectively, while subscripts N and S refers to the north and south channel, respectively; see Fig. 7 for specific locations. (b) 380
Scatterplot of the in-stream contaminant mass discharge at T4 for cDCE and VC with respect to the stream flow rates in the north 381
and south channel, and associated linear regressions. 382
Following this analysis, an additional campaign in autumn 2018 was initiated in order to monitor hydraulic 383
heads in the system with a focus on capturing the dynamics associated with source zone B for a two-month 384
period. The hydraulic head is used here as a proxy for the stream flow, the relationship between hydraulic 385
head and stream flow being relatively stable over this limited period of time and at this time of the year. 386
23
During sustained rain events, an increase in the shallow aquifer water table (head) was observed for the 387
northern part of the site. In addition, hydraulic heads increased for the north channel compared to the 388
levels monitored in the shallow aquifer in source zone B and in the south channel (see hydraulic heads in 389
Fig. 7b: h(GWA), h(SWN) compared to h(GWB) and h(SWS)). These results suggest that sustained rain events 390
associated with the site topography (with a slight slope from north to south along the investigated area) 391
could cause a sufficient hydraulic head difference that activates a substantial inter-channel flow. This inter-392
channel flow leads to the mobilization and transport of contamination stemming from source zone B, which 393
is located very close to the bank at the junction of the two channels. This process was only observed for 394
late times of the year, leading to the hypothesis that vegetation on site may be intercepting and transpiring 395
a significant part of the precipitation during the summer months and forcing the stream water to flow 396
towards the shallow groundwater. 397
398
24
Figure 7. Variation of hydraulic heads, h, for the shallow groundwater on the north side (GWA), shallow groundwater close to 399 the junction on the south side (GWB), and stream water level in the north and south channels (SW N/S). a) Monitoring locations; 400 b) time series for GWA (open red circle), GWB (filled red circle), SWN (filled blue triangle) and SWs (open blue triangle), 401 displayed as a rolling average over 24 hours. Precipitation levels are displayed on the right-hand y-axis (black solid line) 402 corresponding to the black lines at the top of the graph. The black arrows indicate sustained rain events corresponding to a 403 marked increase in GWA; c) Visualization of the water head variations and enhanced hydraulic gradient between the north 404 channel, south channel and GWB before (Time 1) and after (Time 2) a sustained rain event (as indicated by the dotted lines 405 labeled Time 1 and 2). 406
5 Practical implications and perspectives 407
Our study investigated the mass discharge of CAH from a former industrial site to a peri-urban stream and 408
monitored its temporal variations for over a year on a monthly basis. The use of in-stream CMD for the 409
quantification of the mass discharge through strategically placed control transects appeared to be fruitful. 410
It allowed us to track and highlight the contribution of different sources to a receiving stream, while 411
providing valuable insights on the possible contaminant pathways. Furthermore, this approach is valid 412
without consideration of the full mixing of the contaminant, i.e. independently of any prior knowledge on 413
the discharge and mixing zone. It is also a convenient way to aggregate the different pathways from a single 414
source in order to quantify multiple pathway contributions at a larger scale, as carried out with respect to 415
source zone A in this study. This in-stream approach is limited, however, by practical constraints for the 416
discretization of an in-stream control plane, and to dissolved contaminants with concentration levels high 417
enough to be detected after dilution in the stream (Sonne et al. 2018). We observed a substantial increase 418
in CMD related to increases in stream flow. This is in contrast to, for instance, the study by Rønde et al. 419
(2017), who observed an almost constant CMD with time in the Grindsted stream (Denmark). The finding in 420
our study thus implies that the highest concentration in the stream is not necessarily related to the lowest 421
stream flow (largest dilutions), as intuitively expected from equation (2). 422
Our results highlight the importance of understanding the dynamics associated with the presence of 423
multiple contaminant discharge pathways emanating from a contaminant source in peri-urban/urban 424
settings. As expected, shallow groundwater seeping through the streambed played a role, but also key 425
urban features such as former drains and/or sewer lines which acted as preferential flow paths exhibiting 426
different temporal dynamics. Such flow paths were also reported by Rønde et al. (2017) identifying 427
25
drainage culverts as a source of uncertainty in their total CMD calculations. Peri-urban stream systems with 428
such markedly different flow paths can be described as “urban karst”, i.e. comprised of different 429
environmental and engineered compartments as also discussed in Zoboli et al. (2019), and constitute a 430
significant contaminant transport vector for shallow groundwater systems in these settings (Kaushal and 431
Belt, 2012). 432
However, these different pathways were not the main cause for the overall CMD variations documented 433
here, as the largest in-stream CMD downstream did not coincide with the maximum discharge from source 434
zone A. We suggest instead that an inter-channel flow driven by local variations of river stage resulted in a 435
complex contaminant transport from source zone B, located at the junction of two channels, which 436
otherwise seems to be small. Such flow dynamics and resulting contaminant transport is closely related to 437
the meander-driven hyporheic exchange, or transient hydrological conditions at the channel junction scale 438
as described and modelled for example by Boano et al., (2006), Dwivedi et al. (2018) and Han and Endreny 439
(2013). To date, we could not find many other published studies describing point pollutant source dynamics 440
with a source located in the near vicinity of a stream junction (but see Fryar et al (2000) for discussion on 441
the influence of tributary flow and interaction with a contaminant plume). Nevertheless, such a 442
configuration is certainly not unique as many contaminated sites are often historically located in the vicinity 443
of stream waters (Weatherill et al. 2014). 444
The size of the stream and the heterogeneous clayey till setting may also play an important role in the 445
strong CMD variability observed and the resulting variations in concentration levels. Indeed for a given 446
geological setting, small or headwater streams are influenced by local flow systems with high seasonal 447
variations, while higher-order streams and rivers are usually fed by more regional, sustained and steady 448
groundwater components (Winter et al., 1998; Dahl et al., 2007). Furthermore, the low water flow in small 449
stream systems results in a more limited dilution for any contaminant that enters, which in turn has a 450
strong effect on the resulting contaminant concentrations, all CMD variations considered. 451
26
The outcomes of this study can already be employed to facilitate the design of monitoring plan for the 452
assessment of chemical status of streams affected by contaminated sites. It is clear that for some stream 453
systems, the combination of concentration levels and dilution effects without considering the dynamics of 454
any sources present may be misleading. In Denmark for example, the EPA recommends sampling during 455
low flow periods (where highest contaminant concentrations would be expected) while still acknowledging 456
the possible variations of CMD especially in small stream systems (Miljøstyrelsen (Danish EPA) 2018). This 457
study has shown that such a screening approach is only valid when the discharge of contaminants and 458
stream waters are not strongly correlated, i.e. the CMD variations are negligible compared to the stream 459
flow variation diluting the contaminant. Additionally, the sampling frequency was found to be a key factor 460
to consider when designing monitoring strategies for chemical substances. European legislation, e.g. the 461
Water Framework Directive, for the assessment of chemical status has left it open to its Member States to 462
choose what they think is appropriate, although the seasonal variation should ideally be accounted for 463
(European Commission 2009), especially for contaminants with suspected seasonal patterns (e.g. spraying 464
season of pesticides, or tourist-borne substances such as Personal Care Products). 465
6 Conclusions 466
We investigated the temporal and spatial variation of chlorinated ethenes discharging from a contaminated 467
site to a peri-urban stream comprised of two channels. An in-stream CMD approach was applied in order to 468
track these variations at different transects along the investigated stream stretch (ca. 500 m). Our study 469
revealed substantial local variations in concentrations, induced by a highly dynamic contaminant mass 470
discharging to the stream via different pathways comprising a complex system of interlinked environmental 471
and engineered compartments, or urban karst. The in-stream CMD estimates indicated a surprisingly high 472
variation of CMD values, ranging from 1-13 kg/yr depending on the considered transect and measurement 473
period. 474
27
Variable contributions stemming from the groundwater seepage emanating from the complex geological 475
setting and urban drain features were identified in the channelized part of the stream, originating from 476
source zone A. However, these variations alone could not explain the maximum contaminant discharge 477
estimated further downstream. A cross-correlation between different hydrological parameters and the 478
estimated mass discharge revealed a strong link between these quantities and the flow rates in the two 479
channels. Additional hydraulic head measurements suggested that, in periods of sustained rain, a transient 480
hyporheic flow at the junction scale occurs enhancing contaminant transport from source zone B, a second 481
source located at the confluence. Thus, an in-stream CMD approach was found to be an effective method 482
for quantitatively integrating the multiple and highly variable discharge contributions, even if not fully 483
mixed, although information on specific pathways is lost. 484
This study highlights the complexity and variability of contaminant fluxes occurring at the interface 485
between groundwater and peri-urban streams. Notably, small streams are fed by local water flow systems 486
likely to interact with contaminated sites often located in the near vicinity. Consideration of these temporal 487
variations are therefore essential when designing monitoring programs to determine the potential impact 488
arising from contaminated sites impacting streams (for prioritization purposes). Moreover, spatial aspects 489
related to e.g. the presence of multiple active pathways, possibly dependent on the temporal dynamics, 490
become important when considering remedial actions for prioritized contaminated sites in such peri-491
urban/urban stream systems. 492
Acknowledgements 493
This study was partially supported by the Danish Environmental Protection Agency and the Capital Region 494
of Denmark, as well as by a PhD grant from the Department of Environmental Engineering, Technical 495
University of Denmark. 496
497
REFERENCES 498
28
Aisopou, A., P. L. Bjerg, A. T. Sonne, N. Balbarini, L. Rosenberg, and P. J. Binning. 2015. “Dilution and 499
Volatilization of Groundwater Contaminant Discharges in Streams.” Journal of Contaminant Hydrology 500
172: 71–83. 501
Allen, A. 2003. “Environmental Planning and Management of the Peri-Urban Interface: Perspectives on an 502
Emerging Field.” Environment and Urbanization 15 (1): 135–48. 503
Basu, N. B., P. S. C. Rao, I. C. Poyer, M. D. Annable, and K. Hatfield. 2006. “Flux-Based Assessment at a 504
Manufacturing Site Contaminated with Trichloroethylene.” Journal of Contaminant Hydrology 86 (1–505
2): 105–27. 506
Boano, F., C. Camporeale, R. Revelli, and L. Ridolfi. 2006. “Sinuosity-Driven Hyporheic Exchange in 507
Meandering Rivers.” Geophysical Research Letters 33 (18): 1–4. 508
Brack, W., V. Dulio, M. Ågerstrand, I. Allan, R. Altenburger, M. Brinkmann, D. Bunke, et al. 2017. “Towards 509
the Review of the European Union Water Framework Management of Chemical Contamination in 510
European Surface Water Resources.” Science of the Total Environment 576: 720–37. 511
Braud, I., T. D. Fletcher, and H. Andrieu. 2013. “Hydrology of Peri-Urban Catchments: Processes and 512
Modelling.” Journal of Hydrology 485: 1–4. 513
Carter, J. G. 2007. “Spatial Planning, Water and the Water Framework Directive: Insights from Theory and 514
Practice.” The Geographical Journal 173 (4): 330–42. 515
Conant, B., J. A. Cherry, and R. W. Gillham. 2004. “A PCE Groundwater Plume Discharging to a River: 516
Influence of the Streambed and near-River Zone on Contaminant Distributions.” Journal of 517
Contaminant Hydrology 73 (1–4): 249–79. 518
Chapman, S. W., B. L. Parker, J. A. Cherry, R. Aravena, and D. Hunkeler. 2007. Groundwater-surface water 519
interaction and its role on TCE groundwater plume attenuation. Journal of Contaminant Hydrology, 520
91(3-4) 521
29
Dahl, M., B. Nilsson, J. H. Langhoff, and J. C. Refsgaard. 2007. “Review of Classification Systems and New 522
Multi-Scale Typology of Groundwater-Surface Water Interaction.” Journal of Hydrology 344 (1–2): 1–523
16. 524
Danish meteorological institute (Website). http://svk.dmi.dk/dmi/RainEvents/*.login. Last accessed: 525
September 2019 526
Danmark Miljøportal (Website). 527
https://arealinformation.miljoeportal.dk/html5/index.html?viewer=distribution. 528
Last accessed: September 2019 529
Dwivedi, D., C. I. Steefel, B. Arora, M. Newcomer, J. D. Moulton, B. Dafflon, B. Faybishenko, et al. 2018. 530
“Geochemical Exports to River From the Intrameander Hyporheic Zone Under Transient Hydrologic 531
Conditions: East River Mountainous Watershed, Colorado.” Water Resources Research 54 (10): 8456–532
77. 533
Eakin, H., A. M. Lerner, and F. Murtinho. 2010. “Adaptive Capacity in Evolving Peri-Urban Spaces: Responses 534
to Flood Risk in the Upper Lerma River Valley, Mexico.” Global Environmental Change 20 (1): 14–22. 535
Ellis, P. A., and M. O. Rivett. 2007. “Assessing the Impact of VOC-Contaminated Groundwater on Surface 536
Water at the City Scale.” Journal of Contaminant Hydrology 91 (1–2): 107–27. 537
European Commission. 2009. Common Implementation Strategy for the Water Framework Directive 538
(2000/60/EC). Guidance Document No.19: Guidance on Surface Water Chemical Monitoring under the 539
Water Framework Directive. 540
Fisher, H. B. 1979. Mixing in Inland and Coastal Waters. Academic Press. 541
Freitas, J. G., M. O. Rivett, R. S. Roche, M. Durrant, C. Walker, and J. H. Tellam. 2015. “Heterogeneous 542
Hyporheic Zone Dechlorination of a TCE Groundwater Plume Discharging to an Urban River Reach.” 543
Science of the Total Environment 505: 236–52. 544
30
Fryar, A. E., E.J. Wallin, and D. L. Brown. 2000. “Spatial and Temporal Variability in Seepage between a 545
Contaminated Aquifer and Tributaries to the Ohio River.” Groundwater Monitoring & Remediation 20: 546
129–46. 547
Han, B., and T. A. Endreny. 2013. “Spatial and Temporal Intensification of Lateral Hyporheic Flux in 548
Narrowing Intra-Meander Zones.” Hydrological Processes 27 (7): 989–94. 549
Hatt, B. E., T. D. Fletcher, C. J. Walsh, and S. L. Taylor. 2004. “The Influence of Urban Density and Drainage 550
Infrastructure on the Concentrations and Loads of Pollutants in Small Streams.” Environmental 551
Management 34 (1): 112–24. 552
Heejung, K.; H. Hemond. 1998. “NATURAL DISCHARGE OF VOLATILE ORGANIC COMPOUNDS FROM 553
CONTAMINATED AQUIFER TO SURFACE WATERS.” Journal of Environmental Engineering-Asce 124 (2): 554
1998. 555
ITRC. 2010. “Use and Measurement of Mass Flux and Mass Discharge. Interstate Technology & Regulatory 556
Council, Integrated DNAPL Site Strategy Team.,” 1–80. 557
Kaushal, S. S., and K. T. Belt. 2012. “The Urban Watershed Continuum: Evolving Spatial and Temporal 558
Dimensions.” Urban Ecosystems 15 (2): 409–35. 559
Larocque, M., A. Mangin, M. Razack, and O. Banton. 1998. “Contribution of Correlation and Spectral 560
Analyses to the Regional Study of a Large Karst Aquifer (Charente, France).” Journal of Hydrology 205 561
(3–4): 217–31. 562
LaSage, D. M., A. E. Fryar, A. Mukherjee, N. C. Sturchio, and L. J. Heraty. 2008. “Groundwater-Derived 563
Contaminant Fluxes along a Channelized Coastal Plain Stream.” Journal of Hydrology 360 (1–4): 265–564
80. 565
LaSage, D. M., J. L. Sexton, A. Mukherjee, A. E. Fryar, and S. F. Greb. 2008. “Groundwater Discharge along a 566
Channelized Coastal Plain Stream.” Journal of Hydrology 360 (1–4): 252–64. 567
31
Lemaire, G. 2016. “Seasonal Variation of Contaminant Discharge to Stream and In-Mixing Effect.” Master’s 568
Thesis. DTU, Lyngby, Denmark. 569
McKnight, U. S., S. G. Funder, J. J. Rasmussen, M. Finkel, P. J. Binning, and P. L. Bjerg. 2010. “An Integrated 570
Model for Assessing the Risk of TCE Groundwater Contamination to Human Receptors and Surface 571
Water Ecosystems.” Ecological Engineering 36 (9): 1126–37. 572
McKnight, U. S., J. J. Rasmussen, B. Kronvang, P. L. Bjerg, and P. J. Binning. 2012. “Integrated Assessment of 573
the Impact of Chemical Stressors on Surface Water Ecosystems.” Science of the Total Environment 574
427–428: 319–31. 575
Miljøstyrelsen (Danish EPA). 2018. Vandløb Påvirket Af Jordforurening - Tidslig Variation i Koncentration Og 576
Vandføring. 577
Milosevic, N., N. I. Thomsen, R. K. Juhler, H. J. Albrechtsen, and P. L. Bjerg. 2012. “Identification of Discharge 578
Zones and Quantification of Contaminant Mass Discharges into a Local Stream from a Landfill in a 579
Heterogeneous Geologic Setting.” Journal of Hydrology 446–447: 13–23. 580
Niras. 2012. Videregående undersøgelser på Raadvad knivfabrik 3-19 OG 4-50, 581
2800 KGS. LYNGBY. Consulting report for Capital Region of Denmark
Orbicon. 2016. Test af screeningsværktøjet for overfladevand. 582
https://www2.mst.dk/Udgiv/publikationer/2016/04/978-87-93435-60-5.pdf. Last accessed: 583
September 2019. 584
Paul, M. J., and J. L Meyer. 2001. “Streams in the Urban Landscape.” Annual Review of Ecology and 585
Systematics 32 (1): 333–65. 586
Rasmussen, J. J., U. S. McKnight, A. T. Sonne, P. Wiberg-Larsen, and P. L. Bjerg. 2016. “Legacy of a Chemical 587
Factory Site: Contaminated Groundwater Impacts Stream Macroinvertebrates.” Archives of 588
Environmental Contamination and Toxicology 70 (2): 219–30. 589
32
Reid, A. J., A. K. Carlson, I. F. Creed, E. J. Eliason, P. A. Gell, P. T. J. Johnson, K. A. Kidd, et al. 2018. “Emerging 590
Threats and Persistent Conservation Challenges for Freshwater Biodiversity.” Biological Reviews 94: 591
849–73. 592
Rivett, M. O., R. J. Turner, P. Glibbery, and M. O. Cuthbert. 2012. “The Legacy of Chlorinated Solvents in the 593
Birmingham Aquifer, UK: Observations Spanning Three Decades and the Challenge of Future Urban 594
Groundwater Development.” Journal of Contaminant Hydrology 140–141: 107–23. 595
Rønde, V. K. 2014. “Quantification of Stream-Aquifer Contaminant Mass Discharge Using Point-Velocity 596
Probes (PVP) at the Stream Reach Scale (Grindsted, Denmark).” Master’s Thesis. DTU, Lyngby, 597
Denmark. 598
Rønde, V. K., U. S. Mcknight, J. F. Devlin, and P. L. Bjerg. 2017. “Contaminant Mass Discharge to Streams : 599
Comparing Direct Groundwater Velocity Measurements and Multi-Level Groundwater Sampling with 600
an in- Stream Approach.” Journal of Contaminant Hydrology, no. September 2016: 1–40. 601
Roy, J. W., and G. Bickerton. 2012. “Toxic Groundwater Contaminants: An Overlooked Contributor to Urban 602
Stream Syndrome?” Environmental Science and Technology 46 (2): 729–36. 603
Roy, J. W., L. Grapentine, and G. Bickerton. 2016. “Ecological Effects from Groundwater Contaminated by 604
Volatile Organic Compounds on an Urban Stream’s Benthic Ecosystem.” Limnologica. 605
Shepherd, K. A., P.l A. Ellis, and M. O. Rivett. 2006. “Integrated Understanding of Urban Land, Groundwater, 606
Baseflow and Surface-Water Quality-The City of Birmingham, UK.” Science of the Total Environment 607
360 (1–3): 180–95. 608
Simon, D. 2008. “Urban Environments: Issues on the Peri-Urban Fringe.” Annual Review of Environment and 609
Resources 33 (1): 167–85. 610
Skeffington, R. A., S. J. Halliday, A. J. Wade, M. J. Bowes, and M. Loewenthal. 2015. “Using High-Frequency 611
Water Quality Data to Assess Sampling Strategies for the EU Water Framework Directive.” Hydrology 612
33
and Earth System Sciences 19 (5): 2491–2504. 613
Sonne, A. T., J. J. Rasmussen, S. Höss, W. Traunspurger, P. L. Bjerg, and U. S. Mcknight. 2018. “Linking 614
Ecological Health to Co-Occurring Organic and Inorganic Chemical Stressors in a Groundwater-Fed 615
Stream System.” Science of The Total Environment in press: 1153–62. 616
Sonne, A. T., U. S. McKnight, V. K. Rønde, and P. L. Bjerg. 2017. “Assessing the Chemical Contamination 617
Dynamics in a Mixed Land Use Stream System.” Water Research 125: 141–51. 618
Strayer, D. L., and D. Dudgeon. 2010. “Freshwater Biodiversity Conservation: Recent Progress and Future 619
Challenges.” Journal of the North American Benthological Society 29 (1): 344–58. 620
Troldborg, M., W. Nowak, N. Tuxen, P.l L. Bjerg, R. Helmig, and P. J. Binning. 2010. “Uncertainty Evaluation 621
of Mass Discharge Estimates from a Contaminated Site Using a Fully Bayesian Framework.” Water 622
Resources Research 46 (1): 1–19. 623
Vörösmarty, C. J., P. B. McIntyre, M. O. Gessner, D. Dudgeon, A. Prusevich, P. Green, S. Glidden, et al. 2010. 624
“Global Threats to Human Water Security and River Biodiversity.” Nature 467 (7315): 555–61. 625
Walsh, C.J., A. H. Roy, J. W. Feminella, P. D. Cottingham, P. M. Groffman, and R. P. Morgan II. 2016. “The 626
Urban Stream Syndrome : Current Knowledge and the Search for a Cure Source : Journal of the North 627
American Benthological Society , Vol . 24 , No . 3 ( Sep ., 2005 ), Published by : The University of 628
Chicago Press on Behalf of the Society for Freshwate” 24 (3): 706–23. 629
Weatherill, J. J., S. Atashgahi, U. Schneidewind, S. Krause, S. Ullah, N. Cassidy, and M. O. Rivett. 2018. 630
“Natural Attenuation of Chlorinated Ethenes in Hyporheic Zones: A Review of Key Biogeochemical 631
Processes and in-Situ Transformation Potential.” Water Research 128 (May): 362–82. 632
Weatherill, J., S. Krause, K. Voyce, F. Drijfhout, A. Levy, and N. Cassidy. 2014. “Nested Monitoring 633
Approaches to Delineate Groundwater Trichloroethene Discharge to a UK Lowland Stream at Multiple 634
Spatial Scales.” Journal of Contaminant Hydrology 158: 38–54. 635
34
Winter, T. C., J. W. Harvey, O. L. Franke, and W. M. Alley. 1998. “Ground Water and Surface Water: A Single 636
Resource.” USGS Publications, 79. 637
Zoboli, O., M. Clara, O. Gabriel, C. Scheffknecht, M. Humer, H. Brielmann, S. Kulcsar, et al. 2019. 638
“Occurrence and Levels of Micropollutants across Environmental and Engineered Compartments in 639
Austria.” Journal of Environmental Management 232 (October 2018): 636–53. 640
641
642
35
643
FIGURE CAPTIONS 644
Figure 1. Overview of the Raadvad site showing land use, stream flow direction and location of the measurement transects. The dashed red lines correspond to transect locations where detailed stream water sampling was carried out (5 points per transect); solid red lines show the transects used for the in-stream contaminant mass discharge estimation. The dashed black line box indicates the location of the north channel presented in more detail in Figure 2.
Figure 2: (a) Locations for stream water (blue circles) and streambed (piezometers; open squares) samples used in the estimation of contaminant mass discharge (CMD). Approximate delineation (sum of CAHs concentration > 10 µg/L in shallow groundwater) for the main contaminant source zones A and B are marked with striped red polygons based on previous investigations by a consulting company (Niras 2012). Note that the other transects used for CMD estimation are shown in Fig. 1. (b) Pictures from the site indicate the location of the crack and urban drain in the retainment wall.
Figure 3. Effects of mixing and dilution of the in-stream contaminant concentrations measured at 5 detailed transects for TCE and detectable degradation products (cDCE and VC) in June 2016. Note that the x-axis corresponds to a normalized stream width, i.e. 0 for the north bank, and 1 for the south bank.
36
Figure 4. Variations of chlorinated ethene concentrations in the groundwater monitoring well, streambed, and stream in the north channel. (a) Spatial variations for a selected period (April 2017) expressed in PCE eq. (b) Temporal variations of molar ratios for the stream water and streambed water samples at different locations for three selected periods (April, June and November 2017).
Figure 5: Temporal variations of in-stream contaminant mass discharge for chlorinated compounds (TCE, cDCE and VC) along the north channel (T1) and downstream transects (T4, T6), and in the south channel (T7), expressed as PCE eq. See section 3.5 for calculations.
Figure 6. (a) Spearman rank correlation, 𝝆𝝆𝒔𝒔, diagonal matrix between in-stream contaminant mass discharge in transect T4 and selected hydrological parameters: precipitation (N), stream flow rates (Q), water table in shallow aquifer (hGW) at source zone A and B, as well as stream water table (hSW) and the hydraulic gradient (Δh). Subscripts A and B refer to source zones A and B, respectively, while subscripts N and S refers to the north and south channel, respectively; see Fig. 7 for specific locations. (b) Scatterplot of the in-stream contaminant mass discharge at T4 for cDCE and VC with respect to the stream flow rates in the north and south channel, and associated linear regressions.
37
Figure 7. Variation of hydraulic heads, h, for the shallow groundwater on the north side (GWA), shallow groundwater close to the junction on the south side (GWB), and stream water level in the north and south channels (SW N/S). a) Monitoring locations; b) time series for GWA (open red circle), GWB (filled red circle), SWN (filled blue triangle) and SWs (open blue triangle), displayed as a rolling average over 24 hours. Precipitation levels are displayed on the right-hand y-axis (black solid line) corresponding to the black lines at the top of the graph. The black arrows indicate sustained rain events corresponding to a marked increase in GWA; c) Visualization of the water head variations and enhanced hydraulic gradient between the north channel, south channel and GWB before (Time 1) and after (Time 2) a sustained rain event (as indicated by the dotted lines labeled Time 1 and 2).
645
38
646
